Sometimes, images with seemingly clear and obvious plates are missed by our CarCheck API. When this happens, the "results" property of the returned JSON will be empty. If the picture was taken from a mobile phone, your issue may be a rotated image.  Mobile phones can store images with a different rotation than you see on the screen. This is based on how the EXIF headers of the JPEG image are set by the phone's gyroscope. Rotated images will usually not be recognized by our API since the license plate is sideways. 

Most image viewers and web browsers automatically rotate images for display based on the EXIF data, so often this issue cannot be identified visually.  However, certain programming libraries can inspect and remove the EXIF data for you. Below is a Python example using the Python Image Library (PIL)


from PIL import Image
import PIL.ExifTags

img_path = '/path/to/img.jpg'
loaded = Image.open(img_path)
tags = {PIL.ExifTags.TAGS[k]: v for k, v in loaded.getexif().items() if k in PIL.ExifTags.TAGS}
orientation = tags.get('Orientation')
if orientation is None:
    print('No EXIF orientation - good to go!')
else:
    print('EXIF orientation %d - stripping tags and re-saving' % orientation)
    loaded.save(img_path)


If you confirm that image rotation is not an issue, please check point 3 in Improving Recognition.